Re: st: sem multiple correlations and factor weights

[William Buchanan <william@williambuchanan.net>]

Hi Dave,

They very well could be, but it isn't exactly clear what table of multiple correlations you are trying to find.  Although there are people on the list who use SAS, and I would guess a couple that use SPSS, it is a safer assumption that others aren't familiar with programs that aren't Stata.  I, for one, wouldn't use SAS or AMOS for SEM outside of Stata and would use Mplus instead.  Why not create some short example of what it is that you are looking for as the end result so it is explicit to others?  I'm not sure what difference you are making between path weight, indicator path loadings, and factor loadings; in most texts I've read a path coefficient and a factor loading are the same thing if they are part of a measurement model.  Being more explicit will likely get you closer to the result you are looking for.

HTH,
Billy

On Sep 26, 2013, at 12:03 PM, Dave Garson <garson@ncsu.edu> wrote:

> Thank you, Billy, but I am still looking for additional tips on my question.
> 
> 1. Of course in referring to SPSS I was referring to Amos. In SAS I was referring to PROC CALIS, but those packages are only relevant as their output of multiple correlations and factor loadings (not scores or path weights) was what sparked my query.
> 
> 2. I was familiar with estat framework, but this is only part of the multiple correlation output, albeit in unsquared form. The estat framework, standardized table gives multiple correlations (after  squaring), but only for variables which are caused by and only caused by a single exogenous variable in the model. Other multiple correlations are not available in this output as far as I can see.
> 
> I realize this can be done in PCA/PFA, but in SAS and SPSS one does not need to go outside the sem environment. I think the answer is that these outputs are just not available in Stat sem, but I was checking.
> 
> Best to all,
> Dave
> 
> 
> 
> On 9/25/2013 2:33 AM, statalist-digest wrote:
>> Date: Tue, 24 Sep 2013 08:49:45 -0500
>> From: William Buchanan<william@williambuchanan.net>
>> Subject: Re: st: sem multiple correlations and factor weights
>> 
>> Maybe you could be a bit more specific regarding what you are looking for.  SPSS, for example, does not fit SEM (although the AMOS program which extends SPSS does).  So SPSS and SAS have very different capacities in that respect.  If you are referring to PCA/PFA then that is altogether different.  If you want the variance covariance matrix after fitting an SEM in Stata then you can retrieve that from the stored matrices after the estimation.  Additionally, if you use -estat framework- after fitting the model you can retrieve the different matrices used in the estimation of the model:
>> 
>> webuse sem_mimic1, clear
>> sem (SubjSES -> s_income s_occpres s_socstat)(SubjSES <- income occpres)
>> estat framework // This will give you the covariances
>> estat framework, standardized  // This will give you the correlations (i.e., standardized covariances)
>> 
>> HTH,
>> Billy
>> 
>> On Sep 24, 2013, at 8:36 AM, Dave Garson<garson@ncsu.edu>  wrote:
>> 
>>> >For SEM, both SAS and SPSS print out a table of multiple correlations and a table of factor weights of indicator variable loadings on latent variables. I could not find these in Stata's sem, nor in the reference guide (though the guide calls indicator paths "loadings" but that is a different meaning from SAS or SPSS).  Tips on getting this output in sem would be appreciated.
>>> >Dave
>>> >
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/